1 F James J. Angel, Ph.D., CFA Associate Professor of Finance Georgetown University1 Mcdonough School of Business Washington

Home , Expecting (Angel)

James J. Angel, Ph.D., CFA Associate Professor of Finance Georgetown University1 McDonough School of Business Washington DC 20057 [email protected] 1 (202) 687-3765 Twitter: @GuFinProf

May 25, 2017

Securities and Exchange Commission 100 F St. NW Washington, DC 20549-9303 [email protected]

Re: Proposed Rule Change to Adopt NYSE Arca Equities Rule 8.900 to Permit Listing and Trading of Managed Portfolio Shares and various Royce “ETF”s

File SR-NYSEARCA-2017-36

Dear Securities and Exchange Commission:

Here are my comments on this proposal:

Summary

 Arbitrage pricing will be far more difficult for these products than for normal ETPs due to their opaque nature and the inability of arbitrageurs to closely monitor execution quality when the so-

1 All opinions are strictly my own and do not necessarily represent those of Georgetown University or anyone else for that matter.

called “Trusted Agents” execute trades on their behalf. The added costs and risks will lead to wider deviations of the market price from the underlying asset value.

 The selective disclosure of portfolio holdings to so-called “Trusted Agents” raises Regulation FD issues as well as Reg SHO issues. The costly compliance burdens and risks of being a Trusted Agent will limit the pool of willing agents, further driving up the cost of their intermediation.

 These products may fare far worse than normal ETPs during times of market disruption.

 The VIIV calculations are dangerously flawed because they rely on sometimes flawed bid-ask quotes.

 IIV data are important to investors and should be made more readily available by including them in the normal quote feeds.

 SEC should take steps to reduce ETP settlement failures by encouraging ETP sponsors to use smaller sized Creation Units, waive creation fees, and lend ETP shares. Similarly, Rule 15c3-3 should be modernized to make it easier to lend fully paid shares.

 These products are very different from regular ETFs and should not be labeled as ETFs.

This proposal is extremely similar to a previous proposal, File SR-NYSEARCA-2016-08, to launch various Managed Portfolio Shares (MPS). MPS are similar to, but very different from normal ETFs. Instead of the typical transparent ETF structure, the contents of a Creation Basket will only be disclosed to a “Trusted Agent.” Those wishing to engage in the arbitrage activities needed to keep the market price of the fund in line with its underlying portfolio value will have to trade blindly through a “Trusted Agent.” This opaque structure creates a number of serious problems.

Arbitrage pricing will be far more difficult for these products than for traditional ETPs.

The funds propose to rely upon an untested arbitrage mechanism to provide liquidity and prices close to the underlying asset values. However, the actual arbitrageurs and liquidity providers will not get to see the daily portfolio holdings. They may place orders to buy and sell the underlying portfolios only with so-called “Trusted Agents” who are privy to the secret portfolio holdings. However, this adds a layer of complexity and cost to the process. The arbitrageurs and liquidity providers will have little way of judging the execution quality of the fills that they receive from the so-called “Trusted Agents.”

Many arbitrageurs and market makers in ETPs indirectly hedge their ETP exposures through other instruments such as futures or even other ETPs instead of trading directly in the underlying securities. They will be unable to do this if they do not know what is inside the portfolio.

The use of “statistical arbitrage” or “stat-arb” is also problematic for these securities. In statistical arbitrage, traders look for securities that generally trade in the same manner. A classic example is the pair of Coke and Pepsi. As both are large global beverage companies, their stocks tend to go up and down together except when there is news about one of the companies. When the stock prices start to diverge, a stat-arb play would be to buy the stock going down and short the stock going up, expecting to profit when the pattern reverses.

Instead of hedging by trading in the underlying securities through the so-called Trusted Agent, an arbitrageur or market maker may try to hedge using stat-arb techniques. However, the proposed products are actively managed portfolios, whose characteristics may change dramatically from one day – or one minute - to the next. Indeed, the changing nature of the portfolios is one of the purported reasons for the need for secrecy. If the portfolios never changed, then even the last quarterly filing of portfolio holdings would be good enough for market making. A security that seemed to be good stat-arb hedge candidate because it was highly correlated in the past might make a very bad hedge if the underlying secret portfolio has changed. A hedge derived from an analysis of yesterday’s trading data could be totally useless for hedging today because today’s portfolio is different.

More difficult arbitrage implies higher transaction costs and higher deviations from the underlying values.

The increased difficulty, complexity, risk, and expense of arbitraging these proposed funds will undoubtedly result in less arbitrage activity. In order for arbs to step in, the security price has to move far enough away from the underlying asset value in order to make it profitable for the arb to act. Higher costs and risks for the arbs mean that prices will have to move further away from their underlying values before arbitrage activity can occur, leading to larger deviations from the underlying value than for traditional ETPs.

These products may fare far worse during times of market disruption than other ETPs.

ETPs experience problems during times of market disruption, such as during the Flash Crash of May 6, 2010 and the volatility of August 24, 2015. In those situations, the volatility in the main markets spilled over into ETPs, causing many arbitrageurs and liquidity providers to stop trading. This led to large dislocations in which many ETPs traded at prices far removed from their underlying asset values.

Given the opacity and complexity of the arbitrage relationship between these proposed ETPs and their underlying portfolios, it stands to reason that these products would be the first ones that the arbitrageurs and liquidity providers stop trading when the next market disruption hits. In a time of market turmoil, the arbs will understand that the fund’s portfolio manager may be changing the portfolio, and that the so- called Trusted Agents through which arbs are forced to trade may not be able to execute trades in a timely manner. This will cause arbs and market makers to step away from the market, allowing prices to deviate

substantially from the values of the invisible underlying assets. This will substantially harm many retail investors who unwittingly trade at such times.

This product raises Regulation FD issues.

The deliberate opaqueness of the proposed product, together with the proposed disclosure of daily fund holdings to Trusted Agents, raises serious Regulation FD issues as well.2 The spirit of Regulation FD is that material nonpublic information should be released equally to all. The selective disclosure of portfolio information only to so-called Trusted Agents who trade on behalf of their AP clients would appear to be a clear violation of the spirit of Regulation FD.

The “Trusted Agents” will have information the general market does not have, and the humans entrusted with this information will be tempted to profit from that information, either by front running fund trades or by passing the information on to their clients and friends. The ongoing difficulties that the SEC has in enforcing insider trading rules make it most unwise to create yet more valuable secret information that could corrupt humans on trading desks.

The confidential accounts create serious Reg SHO issues.

The market makers who would be trading through confidential accounts generally trade in many other ETFs and related securities. At various times they go long or short as market conditions dictate. As they will not have knowledge of what the position is in their confidential account, they will not know if they are net long or net short on any particular security. This may cause them to unknowingly violate Reg SHO by mismarking a sell order as long when it is short or vice versa.

The confidential accounts create operational issues with settlement and risk management.

2 Regulation FD begins to read as follows:

§ 243.100 General rule regarding selective disclosure.

(a) Whenever an issuer, or any person acting on its behalf, discloses any material nonpublic information regarding that issuer or its securities to any person described in paragraph (b)(1) of this section, the issuer shall make public disclosure of that information as provided in § 243.101(e):

(1) Simultaneously, in the case of an intentional disclosure; and (2) Promptly, in the case of a non-intentional disclosure.

(b)(1) Except as provided in paragraph (b)(2) of this section, paragraph (a) of this section shall apply to a disclosure made to any person outside the issuer:

(i) Who is a broker or dealer, or a person associated with a broker or dealer, as those terms are defined in Section 3(a) of the Securities Exchange Act of 1934 (15 U.S.C. 78c(a)); …

As the account is confidential, would its contents be hidden from the firm’s risk management systems and personnel? This is not clear. Any trading firm has to know its position in real time in order to manage risk. It would be reckless for a firm to not know what is in the confidential account. Otherwise there would be too much risk to the firm. The firm would be vulnerable to a rogue trader like the infamous Nick Leeson.3 If the firm does not know what is in the confidential account, a rogue trader could claim to be doing arbitrage while in practice making dangerous bets far beyond what the firm’s risk controls would allow.

If the firm’s risk management does have access to the contents of the “confidential account”, then it would have to have yet more compliance hassles and have to put information barriers in place that would severely limit the people with access to various parts of the risk management system, further hampering the risk management procedures within the firm.

Furthermore, it is not likely that the confidential account will be closed out exactly at the end of each day. It is likely that late-in-the-day orders to buy or sell a Creation Basket will not be completely filled and that positions will have to be carried over to the next day. This means that the end-of-day positions will have to be settled, which means that the settlement information will be available to settlement personnel. This increases the number of people who need to be sworn to secrecy with the associated compliance headaches.

The “Trusted Agents” will have serious compliance burdens and will have to charge accordingly.

The proposal attempts to provide an untested mechanism that would maintain the secrecy of the portfolio while still allowing something approximating arbitrage to occur. In short, arbitrageurs and market makers would not have access to the undisclosed portfolio holdings and would have to trade through the so-called Trusted Agents.

The so-called Trusted Agents will have serious compliance burdens that may cause firms to either avoid the obligation or else charge fees commensurate to the cost – plus a markup. The so-called Trusted Agents have a duty of confidentiality, and thus will have to have policies and procedures in place to make sure that the top-secret portfolio holdings are kept confidential. They will have to set up procedures to receive the secret information, and severely restrict the employees who have access to that data. The rotation and substitution of trading desk personnel will be far more complicated than before. Similar databases of portfolio holdings that are currently public information will have to be segregated from the secret portfolio data, and the access to that data will have to be restricted and all accesses to the data (including backups) recorded and the records maintained for many years. They will have to train their employees on these procedures and document the training. They will have to monitor adherence to these procedures and document this monitoring, and maintain the documentation in an accessible place for many years. Even if the firm has sound policies and procedures, it is likely that eventually someone will

3 Nick Leeson was a trader at Barings in Singapore. While he was purportedly engaged in arbitrage trading, he took very large bets that destroyed Barings when the bets turned sour.

fall for the temptation and seek to profit from this information, leading to expensive enforcement actions against the firm.

The compliance hassles will extend to the operations personnel who have to clear and settle the trades. There will have to be policies and procedures in place to limit who has access to those accounts, and what information can be given even to the customer.

These burdens are likely to drive up the cost of being a Trusted Agent. They will thus have to charge accordingly, driving the cost of arbitrage above the cost of arbitraging traditional ETFs. The higher costs and higher compliance risks will severely limit the number of firms willing to take on the burden of becoming Trusted Agents. With less competition, arbitrageurs and market makers are likely to face even higher fees and poorer service. In the event that there were many Trusted Agents, then the likelihood of data breaches would also increase.

The construction of the VIIV is flawed as it is based on sometimes flawed bid-ask midpoints.

The proposed price feed, the “Verified Intraday Indicative Value” (VIIV), is based on the midpoint of the bid-ask spread. While in theory the bid-ask midpoint may appear to present a better estimate of the true price of a liquid stock than the most recent trade, this is not always the case for all exchange-listed securities that are fair game for these funds. For some relatively illiquid stocks, there are times when the bids and offers are absurd, and thus the bid-ask midpoint is likewise absurd. Even one such stock in a portfolio can severely distort a VIIV based on bid-ask midpoints.

The proposing release argues that bid-ask midpoints are better than stale prices for securities that are thinly traded. This is not always the case, as indicated by the following example.

Suppose that the Generic Example Fund (GEF) has a nicely diversified portfolio of 50 equally weighted stocks. Assume that 49 of the 50 stocks are quoted at $19.99 bid and $20.01 asked, with a bid-ask midpoint of $20.00. However, the 50th stock is Essa Pharmaceuticals (EPIX), which as of 9:30:03 on June 1, 2016 was quoted at $.01 bid, $199,999.00 asked, an actual real-life example. This leads to a midpoint price of $99,999.51 for EPIX. The properly calculated VIIV of GEF based on bid-ask midpoints is thus (49/50)*$20.00 + (1/50)*$99,999.51 or $2,019.59. Alas, EPIX opened at $3.17 and closed at $3.20. My point is that just one bad bid or offer quote can lead to a wildly inaccurate VIIV value. The VIIV should instead be based on the last trade, but if the underlying market is closed or the underlying asset has not traded recently, then a reasonable fair value methodology should be used.

Access to IIVs is important to protect retail investors from mispricing.

Although most ETP prices closely track the values of their underlying portfolios, sometimes they do not. In order to alert investors when this is the case, it is essential that investors have easy access to the

Intraday Indicative Value (IIV) data.4 IIV data can be quite useful in protecting retail investors in circumstances where the ETP price is quite different from the underlying value.

For example, the iPath GSCI Crude Oil Trust (ARCA: OIL) was selling for substantially more than its underlying value for quite some time.5 At times this discrepancy was more than 20%, as seen in the following chart:

Unwitting investors ended up purchasing this ETP at a substantial premium to the underlying value of its assets.

Here is another example. The following is a screen shot of an Interactive Brokers quote screen from 10:15 am on June 6, 2016 showing the last trade and the IIV data for a few ETPs. Note that the NUGT is selling at a slight discount of 0.1% to its IIV. However, the inverse DUST, which is based on the same underlying index of gold mining stocks (most of which are in the US and Canada and trading during U.S. market hours), is selling at an 8.7% discount.

4 Although IIVs are far from perfect and can and should be improved, they are still useful. See my previous comment letter on ETPs. https://www.sec.gov/comments/s7-11-15/s71115-29.pdf.

5 See http://seekingalpha.com/article/3898146-overpricing-oil-etf-presents-arbitrage-opportunity for more details.

This is an arbitrage big enough that even a slow retail investor like me could do it by buying the underpriced DUST and hedging by shorting NUGT. Even if a retail investor is not eager to do the arbitrage, better information about the IIV could warn investors about times when the ETP prices are out of alignment from their underlying portfolio values.

IIVs should be disseminated over the normal quote feeds.

One major defect in the ETP universe is the difficulty most investors have of obtaining real-time IIV data. Simply put, most brokerage firms do not provide this information to their customers.6 Yahoo! Finance (unlike Google Finance) does display IIV data, but one has to know where to look. (To get an IIV on Yahoo, type ^XXX-IV where XXX is the ticker symbol.)

Although ETP sponsors generally display IIV data on their web sites, it is not clear how timely these data are. Even if such data were timely, it is awkward for individual investors to find the data when they want to trade.

My understanding is that the IIV data are not disseminated over the normal quote feeds and that there are several different data feeds with the data. It is quite cumbersome for any entity to gather these different IIV feeds together. They should all be disseminated at no extra cost over the same normal consolidated quote feeds used by the industry for standard NMS stock quotations.

6 Two notable exceptions are Merrill Edge and Interactive Brokers, both of which make IIV data available to investors. I have not been able to find the IIV data on E-trade, TD Ameritrade, Schwab, Vanguard, Fidelity, Scottrade, or finance.google.com.

The SEC should take steps to mitigate ETP settlement failures.

The ongoing large and protracted settlement failures in ETFs are a blemish on the integrity of the U.S. equity markets. Their persistence despite the Regulation SHO Threshold List mechanism and Regulation 204’s knife-edge buy-in requirements are an embarrassment to the U.S. equity markets and its regulator, the SEC. Regulation 204 cleaned up most of the settlement failures in corporate stocks, but the SEC allows the problem to continue to fester in ETPs.

This is ironic in that settlement failures are even more problematic and damaging to ETPs than to corporate stocks. Persistent settlement failures and long-term membership on the Reg SHO Threshold list are not just cosmetic problems. These protracted failures exacerbate the illiquidity facing many ETPs. Stocks experiencing protracted settlement failures and Regulation SHO treatment are hard to borrow, which means that they are difficult to short, and expensive to borrow when they can be shorted. This makes it much more difficult to short ETPs, narrowing the pool of traders willing and able to arbitrage ETPs and their underlying assets.

With fewer traders willing to conduct arbitrage and provide liquidity, there is less liquidity, higher trading costs, and the deviations between ETP prices and the underlying market values are likely to be larger and more frequent.

The higher cost and inability to borrow/short these ETPs shares with high delivery fails increases the risks to investors in short positions. I have personally and painfully been bought in on a short position in an ETF. Even though market makers are not bought in until T+6, they too may be hesitant to provide liquidity due to the risk of being forced to buy in because they could not borrow the shares.7

The NUGT/DUST mispricing described above is a classic example of how the frictions in the stock lending market are harming retail investors who expect ETPs to closely track their underlying values. DUST is seriously mispriced relative to NUGT, which is based on the exact same underlying index of gold mining stocks. One would expect arbitrageurs to step in and purchase DUST and short NUGT. Alas, NUGT is not an easy-to-borrow stock and most retail brokerages would reject an order to sell short. (DUST is even worse and is on the Regulation SHO Threshold List.) Note that it is not the leverage in these products that is leading to the mispricing. There are many other pairs of leveraged products which

7 It is a common misconception that market makers have until T+6 to settle their trades. Regulation 204(a) does not exempt market makers from the obligation to deliver shares on the settlement date. It clearly states “A participant of a registered clearing agency must deliver securities to a registered clearing agency for clearance and settlement on a long or short sale in any equity security by settlement date…” The misconception stems from the buy-in provision of 204(a)(3) which states that the buy-in for a fail position resulting from legitimate market making would be on the third day after the settlement date. Thus, market makers are still required to settle on the regular settlement date, but they do not need to be bought in until three days after the settlement date. Market makers who make practice of not delivering on the regular settlement date may be subject to enforcement action.

do not have this level of mispricing. It is the frictions in the create/redeem process and the stock lending market that are leading to these harmful mispricings.

As it is harder and more expensive to conduct arbitrage on short-constrained ETPs, such ETPs are more likely to experience disruptions during times of market turmoil. With fewer arbitrageurs and liquidity providers able to engage in arbitrage and liquidity provision, during times of market disruption it becomes more likely that there will be no arbitrage and ETP prices will deviate substantially from underlying values, as occurred on August 24, 2015 and in the Flash Crash.

The following steps should be considered as a means of mitigating settlement failures:

 Reduce size of Creation Units. ETP issuers should be encouraged to keep the size of Creation Units, the minimum number of shares that can be created or redeemed, as small as possible, even as low as 100 shares for some ETPs. The only real constraint is a need to avoid fractional shares for a large number of constituents. Indeed, fractional shares for a small number of the smallest constituents need not be a show stopper because such small amounts can be settled in cash. As creation and redemption can only be done by Authorized Participants, there is no danger that large numbers of retail investors will bother creating or redeeming small numbers of shares.

 Encourage ETPs to waive fees for creating ETPs. There is no reason why the fees for creation/redemption need to be symmetrical. Operators of toll bridges that are hard to bypass such as the Golden Gate Bridge long ago figured out that it was much more efficient to charge twice the toll in one direction and not bother to charge in the other direction. ETP sponsors should be encouraged to do the same thing. This is in their best interest as it will encourage the creation of more shares and thus lead to more assets under management. By making it easier and cheaper to create ETP shares, there is less excuse for failing to deliver the shares.

 Modernize Rule 15c3-3 to make it easier to lend fully paid shares out of cash accounts. The well-meaning Customer Protection Rule (15c3-3) imposes numerous paperwork burdens on lending out fully paid shares. However, it is much easier to lend out shares from margin accounts with debit balances. The industry’s long track record of safely lending margined shares while protecting consumers shows that the current rules on lending fully paid shares can be relaxed without harming consumers.

 Permit issuers to lend ETP shares. In a typical stock loan, the lender delivers shares and the borrower puts up cash collateral that is marked to market daily. ETP issuers should be permitted to do effectively the same thing in any amount. They should be permitted to instantly create ETP shares on demand using cash collateral, with the understanding that either the underlying shares will be delivered in a timely manner or the loaned ETP shares will be returned. This could totally eliminate the need felt by some market makers to fail to deliver ETP shares on the standard settlement date. This would be little different than the practice that some ETPs have of permitting cash creation and redemption. 10

 Examine other means of reducing frictions in stock lending. As part of its reviews of market structure, the SEC should also examine securities lending to find other means of reducing unnecessary frictions in the process.

The proposed funds are very different from ETFs and should not be labeled or approved as such.

This proposal is for a different type of product that would be marketed as an ETF. For all of the reasons above, it is likely that these products will suffer from a lack of arbitrage activity, leading to larger than normal discrepancies between the underlying portfolio value and the market price. Retail investors rely upon arbitrage to keep ETF prices properly aligned with the underlying portfolio values. These products should not be approved as ETFs in their present form as investors will be misled. Investors will assume that their market prices will be close to the underlying portfolio values when in practice they will deviate much more substantially than most ETFs.

If the Commission does decide to approve these products, it should require that they be clearly labeled as different from ETFs. The backers should be prohibited from marketing them as ETFs or allowing them to be misrepresented as ETFs. Furthermore, VIIVs should be disseminated over the standard consolidated feeds, not specialized feeds, such that the data are widely available to all investors.

Respectfully submitted,

James J. Angel, Ph.D., CFA Georgetown University